Ellidiss Technologies w w w . e l l i d i s s . c o m Ellidiss - PowerPoint PPT Presentation

LAMP: A new model processing language for AADL ERTS 2020 Toulouse, 30 Jan 2020 P. Dissaux Ellidiss Technologies, 24, quai de la douane, 29200 Brest, France Ellidiss Technologies w w w . e l l i d i s s . c o m

Ellidiss Technologies Model Processing 1/2 w w w . e l l i d i s s . c o m Model processing goals • Observe: model exploration, queries and views • Enrich semantics: add constraints not reflected by the syntax • Verify static properties: rules checkers • Perform model transformations: • Model to Model transformations: • Refinement: transform a descriptive models along the development life-cycle • Verification: transform a descriptive model into a verification model • Model to Text transformations • Code generation • Documentation generation Model processing requirements • Require a model processing language • With a way to get access to model éléments (API) • With strong and well defined semantics • Especially to support verification activities for critical systems (qualification) • Better with an existing set of predefined rules and utilities (libraries)

Ellidiss Technologies Model Processing 2/2 w w w . e l l i d i s s . c o m Model Processing Languages • May be specific to the modeling language or development environment: • OMG world: OCL, ATL, QVT, MOFM2T • AADL world: REAL, LUTE, RESOLUTE, AGREE • Or a general purpose programming language: Java, Ada, Python, Prolog • Declarative style is appropriate to express rules Model Processing Applications • Off-line processing: • Part of a modeling/verification environment (tool plugins) • Under responsability of the tool editor • Independent from the actual model instance that will be processed • On-line processing: • Part of the model to be processed (model extension) • Can be developed/customized by the end user • May access actual model instance information • Can also be used to prototype an off-line processing tool

Ellidiss LMP: Technologies w w w . e l l i d i s s . c o m a generic model processing approach LMP: Logic Model Processing • Using the Prolog language as a Model Processing language (Declarative predicates, Boolean logics, unification and backtracking) • Applying a dedicated development and runtime process: LMP development and runtime process: • Any Meta-Model can be represented by a set of Prolog Fact Definitions • Any Model can be represented by a populated Prolog Facts Base • Any Model processing action (queries, constraints, transformations, …) can be represented by a Prolog Rules Base • Prolog facts and rules base can be merged together to get the expected result LMP Facts Model A Base A Facts Expected Prolog Model B Result Base B Engine Rules Goal X Base X

Ellidiss Technologies LMP runtime process w w w . e l l i d i s s . c o m Model processing Models to be result processed parser or input Facts base parser or prolog generator input Facts base parser or result generator input Facts base engine generator Textual Output Model Input Model or Parsers: Check Report byte code in files - aadlrev Source code or … Rules bases - xmlrev Rules bases in memory Rules bases Rules bases - adarev Rules bases - crev Generators: processing library: - Stood - Model analysis - AADL Inspector - Model exploration - C library - M2M transformation - M2T transformation Model processing applications

Ellidiss Technologies w w w . e l l i d i s s . c o m LMP for tag-based models (XMI/XML) Fragment of the meta-model: XML Schema Definition (XSD) <xsd:schema ... > <xsd:complexType name="SPEC-OBJECT"> ... <xsd:attribute name="DESC" type="xsd:string" use="optional"/> <xsd:attribute name="IDENTIFIER" type="xsd:ID" use="required"/> <xsd:attribute name="LAST-CHANGE" type="xsd:dateTime" use="required"/> <xsd:attribute name="LONG-NAME" type="xsd:string" use="optional"/> </xsd:complexType> ... </xsd:schema> Corresponding prolog Fact Definition: isSpecObject( Desc,Identifier,LastChange,LongName ) .

Ellidiss Technologies w w w . e l l i d i s s . c o m LMP for token based models Fragment of the meta-model: Backus-Naur Form (BNF) … component_type ::= component_category component_identifier { property }* end component_identifier ; property ::= property_name => property_value; … Corresponding prolog Facts definition: isComponentType( ComponentCategory,ComponentIdentifier ) . isProperty( ComponentIdentifier,PropertyName,PropertyValue ) .

Ellidiss Technologies LMP: Merging and Processing w w w . e l l i d i s s . c o m 1. Merge together the two facts bases: model A (requirements) isSpecObject( '','','','Temp_Lower_Bound' ) . isSpecObject( '','','','Temp_Upper_Bound' ) . ... isComponentType ('Thread','Thermostat'). isProperty( 'Thermostat','Coverage','Temp_Lower_Bound' ) . isProperty( 'Thermostat','Coverage','Temp_Upper_Bound' ) . model B 2. Add the rules base: (design) check requirements coverage: checkCoverage :- isSpecObject( _,_,_,R ), rule to isComponentType (_,C), check not (isProperty( C,'Coverage ’,R )), writeErrorMessage( R ). 3. Run the Prolog engine

Ellidiss Technologies w w w . e l l i d i s s . c o m Benefits of the LMP approach • Generic solution for: • Model queries • Model constraints • Model transformations: M to M or M to T • Model exploration and architectural reasoning • Standard prolog language (ISO/IEC 13211-1) • Independent : compatible with the main meta-modelling formats (BNF, XSD, Ecore) • Declarative : rules oriented approach • Modular : • separate fact and rules bases • rules bases transitivity: e.g. Marte to Cheddar : Marte to AADL and AADL to Cheddar • Formal (boolean logic): appropriate for tool qualification • Flexible : • Supports heterogeneous models • Supports incomplete models (subsets) • Supports erroneous models (debugging) • Industrial return of experience of many off-line processing tools: • Airbus: LMP applied for the verification of DO-178 certified projects (A380, A350) • European Space Agency: used in the TASTE tool-chain • Honeywell: architecture reasoning • Ellidiss: AADL Inspector model adaptors and Stood code generators • Commercial support

Ellidiss Technologies w w w . e l l i d i s s . c o m AADL • Architecture Analysis and Design Language: – Describes Systems with Hardware and Software components – Formal static and run-time semantics (real-time) – Textual and graphical notations • SAE aerospace division – AS-2C subcommittee: – AADL 1.0 (AS 5506) 2004 – AADL 2.0 (AS 5506A) 2009 – AADL 2.1 (AS 5506B) 2012 – AADL 2.2 (AS 5506C) 2017 – AADL 3.0 : in preparation • Annex documents – Annex A: ARINC 653 Interface (AS 5506/1A) 2015 – Annex B: Data Modelling (AS 5506/2) 2011 – Annex C: Code Generation Annex (AS 5506/1A) 2015 – Annex D: Behavior Annex (AS 5506/3) 2017 – Annex E: Error Model Annex v2 (AS 5506/1A) 2015 • Online resources – https://www.sae.org/standards/content/as5506c – http://www.openaadl.org

Ellidiss Technologies Off-line LMP plugins w w w . e l l i d i s s . c o m in AADL Inspector LMP plugin category AADL semantic rules model checker AADL instance builder model exploration AADL ARINC 653 rules model checker UML MARTE to AADL model transformation SysML to AADL model transformation Capella to AADL model transformation AADL to Cheddar model transformation AADL to Marzhin model transformation AADL to OpenPSA model transformation AADL printer model unparser LAMP checker model checker

Ellidiss On-line LMP programs for AADL: Technologies w w w . e l l i d i s s . c o m the LAMP annex package Ellidiss::ERTS2020::paper26::e1 public abstract A -- a LAMP annex at component level annex LAMP {** /* standard prolog syntax */ **}; end A; -- a LAMP annex at package level annex LAMP {** /* standard prolog syntax */ **}; end Ellidiss::ERTS2020::paper26::e1;

Ellidiss Technologies The LAMP stack w w w . e l l i d i s s . c o m LAMP Checker AADL Inspector plugin LAMP Local Rules in AADL Components LAMP Rules Libraries in AADL Packages LAMP Std Library LMP Facts in AADL Packages Prolog Libraries LMP Libraries Prolog Engine prolog byte code prolog source code uses

Ellidiss Inside LAMP Technologies w w w . e l l i d i s s . c o m LAMP AADL Annex subclause: • Syntax: ANNEX LAMP {** /* standard prolog syntax */ **}; • LAMP user defined libraries in AADL Packages • LAMP user defined local rules in AADL Components • No new langage to define and maintain • Direct access to the LMP low level API (all AADL model elements) • Can also work on incorrect models (debugging) LAMP standard library: LAMPLib.aadl: • High level API to the AADL declarative model • High level API to the AADL instance model • High level API to the Behavior and Error annexes • API to analysis results (e.g. simulation traces) • Utility rules (printing, …) LAMP support inside AADL Inspector • LAMP checker analysis plugin • LAMPLib is pre-loaded within the AADL "environment" • Available since A.I. 1.7 (http://www.ellidiss.com/downloads)